Welcome to the StaR Child Health Risk of Bias Wiki!
What is StaR Child Health?
StaR Child Health stands for Standards for Research in Child Health. This is a group of methodologists, clinicians, and policy-makers who seek to enhance the quality, ethics, and reliability of pediatric clinical research by promoting the use of evidence-based standards. Within StaR Child, there are a number of Standard Development Groups, including one focused on risk of bias.
This wiki is meant to:
provide information on risk of bias in randomized controlled trial design and conduct
be a source of tools and references that explain and help minimize risk of bias in randomized controlled trials
encourage discussion about pediatric randomized controlled trial design
This wiki is meant for:
pediatric trialists and trainees
researchers designing protocols for randomized controlled trials
researchers interested in learning about bias and methodology in randomized controlled trials, including systematic reviewers
This wiki is currently under evaluation by:
What is risk of bias?
Bias is a systematic error that, when introduced into a study, increases the chances that treatment benefits and/or harms will be over- or underestimated This in turn impacts the validity of the results. There are two types of validity: internal and external. The risk of bias in a study is a measure of its internal validity - the degree to which the study results "should be believed."[Ch. 8.2.2. Higgins] It is not a measure of factors that influence external validity, such as generalizability and precision.
There are seven key domains relevant to risk of bias. These domains tend to occur at different stages of a randomized controlled trial, as illustrated in the figure to the right. The definitions of the domains are given below:
Appropriate sequence generation can minimize selection bias. It is the method of assigning (or allocating) participants to study groups. A random allocation sequence makes it more likely that the groups being compared are balanced for known and unknown confounders.
Appropriate allocation concealment can minimize selection bias. Proper allocation concealment maintains protection of the allocation sequence until the point when group assignments are made. This ensures that the person enrolling participants is unaware of the group assignments to prevent selective enrollment or allocation. This can always be done!
Blinding of participants and study personnel can minimize performance bias. If study participants and key personnel are unaware of the intervention group they belong to, there is less of a chance that their behaviour will be altered due to expectations or beliefs that may be associated with that group.
Blinding of outcome assessment can minimize detection bias. By blinding the individual responsible for assessing the study outcome to the group assignment, there is less of a chance that their assessment or interpretation will be influenced by factors related to exposure status.
Incomplete outcome data can lead to imbalanced comparisons of groups and attrition bias. Important considerations include properly addressing incomplete data, keeping track of attrition and the reasons for it, and using intention-to-treat analysis. These can highlight whether there are systematic differences between groups that might bias the results.
Selective outcome reporting is the reporting of a subset of the originally specified outcomes for the trial, leading to reporting bias. Two factors are important to consider: definition of the primary outcome, which should remain constant from the protocol to the publication, and inclusion of all outcomes in a report, regardless of factors like statistical significance.
Other sources of bias relates to other study features that may introduce bias into a trial. This includes things like baseline imbalances, blocked randomization in unblinded trials, and study design characteristics.
Thanks to our sponsors!
Presentation on risk of bias in pediatric randomized controlled trials, given at the 3rd StaR Child Health Summit in May 2012 in Winnipeg, Manitoba. Watch the video from the meeting here: http://bit.ly/KCQgVj
Welcome to the StaR Child Health Risk of Bias Wiki!
What is StaR Child Health?
StaR Child Health stands for Standards for Research in Child Health. This is a group of methodologists, clinicians, and policy-makers who seek to enhance the quality, ethics, and reliability of pediatric clinical research by promoting the use of evidence-based standards. Within StaR Child, there are a number of Standard Development Groups, including one focused on risk of bias.This wiki is meant to:
This wiki is meant for:
This wiki is currently under evaluation by:
What is risk of bias?
Bias is a systematic error that, when introduced into a study, increases the chances that treatment benefits and/or harms will be over- or underestimated This in turn impacts the validity of the results. There are two types of validity: internal and external. The risk of bias in a study is a measure of its internal validity - the degree to which the study results "should be believed."[Ch. 8.2.2. Higgins] It is not a measure of factors that influence external validity, such as generalizability and precision.There are seven key domains relevant to risk of bias. These domains tend to occur at different stages of a randomized controlled trial, as illustrated in the figure to the right. The definitions of the domains are given below:
Thanks to our sponsors!
Presentation on risk of bias in pediatric randomized controlled trials, given at the 3rd StaR Child Health Summit in May 2012 in Winnipeg, Manitoba. Watch the video from the meeting here: http://bit.ly/KCQgVj